doc_xpath.py

#!/usr/bin/env python
# encoding: utf-8
# author: 04


def doc_xpath(doc, xpath):
    """
    >>> a = {'a': [{'b': {'c': ['d', 'f']}}, {'b': {'c': ['g', 'h']}}]}
    >>> doc_xpath(a, 'a.b')
    Traceback (most recent call last):
       ...
    ValueError: Xpath <a.b> failed at <b>.
    >>> doc_xpath(a, 'a.[].b.c')
    [['d', 'f'], ['g', 'h']]
    >>> doc_xpath(a, 'a.[].b.c.[]')
    ['d', 'f', 'g', 'h']
    """
    check_points = [doc]
    for word in xpath.split('.'):
        new_points = []
        for point in check_points[:]:
            try:
                if word == '[]':
                    if not isinstance(point, list):
                        raise ValueError()
                    new_points.extend(point)
                else:
                    if not point.get(word):
                        raise ValueError()
                    new_points.append(point.get(word))
            except Exception:
                raise ValueError('Xpath <%s> failed at <%s>.' % (xpath, word))
        check_points = new_points
    return check_points


if __name__ == "__main__":
    import doctest
    doctest.testmod()